Evaluating Our Estimates and Projections

At Claritas, we’re often asked how we measure the accuracy of our demographic products. Our demographers, Ken Hodges and Rachel Cortes, recently wrote a paper to answer that very question. Both Ken and Rachel bring a wealth of knowledge and experience to the Claritas demographic program. Ken has a Ph.D. in Demography from Cornell University and over 35 years of experience in private-sector demography. He is very active in the applied demography community, acting as a board member of the Council of Professional Associations on Federal Statistics. He’s also a member of other groups such as the Population Association of America Committee on Population Statistics and the Census Data Products Redesign Group. Ken even presented at the Workshop on 2020 Census Data Products: Data Needs and Privacy Considerations.

Prior to joining Claritas, Rachel previously worked at the Census Bureau, the Population Reference Bureau, and The San Diego Association of Governments. She is also very active in the applied demography community acting as a member of the Population Association of America. Plus, she served on the Council of Professional Associations on Federal Statistics’ Blended Base and Data Quality Subcommittees. Rachel was also a presenter at the 2020 Census Data Products: Public Workshop on the Demographic and Housing Characteristics Files.

Why Is Census Data Important To Marketers?

When looking for Census data you can make actionable business decisions with, you come to a provider like Claritas. We use Census data to help inform businesses on where to expand operations, what products or services to offer to their customers, and even what consumer segments exhibit specific behaviors, for added marketing guidance.

Having access to accurate counts, estimates, and projections of individuals and households is critical. Knowing that the data we provide closely aligns with what the U.S. Census Bureau releases every 10 years provides our clients with peace of mind. They rely on us to help them stand out in a crowded marketplace and ensure they invest their marketing dollars where it will make the most impact.

Why Do We Measure Against the 10 Year Census?

In addition to knowing who our experts are, it’s also important to note why we measure against the decennial (ten-year) Census. The closer we get to the next Census, the more requests Claritas receives for more recent comparisons. That said, the decennial Census is truly the only good opportunity to undertake this measurement. There’s simply nothing else as rigorously comprehensive or representative. While other surveys and data sources are incredibly valuable, nothing else comes close to measuring the entire United States population. That’s why the Federal Government is the only entity with both the resources and authority to take on such an enormous task. Despite that, they only attempt it once a decade! Even the Census Bureau’s own population estimates that are done in between decennial Census updates are based on administrative data and are subject to error.

If you’ve been involved with private sector demography for any length of time, you may have heard demographics referred to as a commodity. They’re kind of like burgers. However, there’s still a major difference between fast food and gourmet. At Claritas, we pride ourselves in our ingredients, our method, and our team. Our paper evaluating our estimates and projections is evidence of our dedication to creating a gourmet product.

Accurate Results

The Claritas 2020 population and household estimates achieved impressive levels of accuracy and comparisons with previous evaluations indicate continued improvement over the decades. The finding of higher error for rapidly changing areas has become so standard as to almost go without reporting. However, despite concerted efforts to reduce estimation error in rapid change areas, the improved accuracy has been primarily in stable and moderately changing areas, where one might expect improvement to be more difficult to achieve. Still, it is where change has been greatest that the 2020 Claritas estimates provided the most frequent improvement over the use of previous Census counts. In other words, it is in rapid change areas, with some of the larger estimation errors, that the Claritas estimates delivered the greatest value to data users.

Another rule of thumb is that estimation error is larger for areas with smaller population or household totals. This was certainly the case across geographic levels. Block group error was larger than tract error, which was larger than county error, and so on. Within small geographic levels, such as block groups, size and error were not always strongly related. Mean percent error was often higher for areas with larger totals. This may be because some larger totals were reached through the kind of rapid growth associated with higher estimation error.

Limited Bias

We’re happy to report that the Claritas estimates also achieved low levels of bias. This is important since applications usually involve aggregations to user-defined areas. With low bias, both accuracy and improvement over previous Census counts are usually enhanced. This includes aggregations of just a few block groups.

Quality Projections

Long term projections are always tricky. This is because five years into the future it is harder to predict change to persons and households, especially at a small geographic level. But, the 2020 Claritas projections had only slightly more error than the estimates when compared to the 2020 Census.

Significant Value

All small area population and household estimates are subject to error, and must be used with care. That said, the Claritas estimates have demonstrated impressive accuracy, and steady improvement over the decades. There is ample room for improvement, but this evaluation confirms our estimates provide significant value for Claritas data users.

Learn More

For a copy of Claritas Evaluation of 2020 Estimates Against 2020 Census Results, fill out our ‘Contact Us‘ form. One of our dedicated team members will be in touch.